LangWatch’s Post

Huge shout out to our dev team for making this happen over the past few weeks. Through close collaboration with our customers, we successfully shipped these crucial features on our platform. Check the comments section for the video links from the slides.

We are thrilled to share that LangWatch is the official judge for the Langflow (by DataStax) Hackathon Competition! 🎉 This hackathon, focused on their large Brazilian community, is an incredible opportunity for the 3500 AI developers to showcase their skills in developing the best RAG applications. Our platform's evaluators will be diving into the quality of these RAG applications to pick the ultimate winner. The competition kicks off today, and we can't wait to see the creativity and expertise in action! 🤩 This is a fantastic way for us to showcase the best methods for evaluating LLM quality. Good luck to all participants! 🔥 #LangwatchAI #LangflowHackathon #Innovation #AI #LLM #Hackathon

🚀 Yesterday, Rogério Chaves took the stage at AI Builders (Join this community if you haven't yet) to speak about a more and more popular topic: Automated prompt engineering! 👏 Kudos to Rogerio for an incredible session 🌟 He showcased optimizing a RAG solution and visualized the results using the DSPy framework and visualizer. Interested in learning more about this topic and missed the session? Reach out to us via: https://lnkd.in/eN_PpebD and mention "auto-prompting" in the comments. #AI #DSPy #AutomatedPromptEngineering #Innovation

🚀 Exciting News for LLM Enthusiasts! 🚀 Have you started building your LLM-powered solutions with FlowiseAI (YC S23) We’re excited to share that LangWatch is now fully integrated into Flowise! 🌟 Build your perfect AI solution while we handle monitoring, quality checks, and provide full visibility into your user analytics. Say goodbye to the black-box and gain deeper insights into your AI's performance. #AI #LLM #genai #llmops #monitoring #qualitycontrol #UserInsights

Excited to share that LangWatch is at The AI Summit in London today and tomorrow! 🇬🇧 Visit us at the Dutch booth, "The Garden," where we’ll be joined by other amazing startups. Don't miss the opportunity to stop by and meet with us! Thanks to Rijksdienst voor Ondernemend Nederland (RVO) for the organisation.

"After many hours of manual prompt engineering we couldn't get the accuracy higher than 60%, DSPy took it to 80% for us in a matter of minutes." DSPy is the hot new framework on LLM community, which takes a different approach than what so far has been done for the past year: instead of having engineers manually trying to adjust prompt to improve results and do a *vibe-checking* on the outputs. That's why LangWatch implemented the DSPy Visualizer, which allows you to log your DSPy training sessions, track the performance, costs, compare runs and debug them in detail on the LangWatch platform. Read more here: https://lnkd.in/e_e8e_m4 Curious to try it out? Book a demo with us: https://lnkd.in/eN_PpebD #dspy #promptengineering #llms #genai

Preventing 'bullshit' output in LLM-powered applications is crucial for ensuring reliable and trustworthy AI interactions. Setting up the right quality evaluations is key to mitigating this issue. Here are some tips to get started: 1. Establish Baselines: Define clear baselines for what constitutes acceptable and unacceptable responses. This helps in identifying deviations more effectively. 2. Context Relevancy: Evaluate whether the AI’s responses are contextually relevant. This means checking if the information provided aligns with the query and context given. 3. Faithfulness: Ensure that the responses are faithful to the provided context and do not introduce information that cannot be inferred from it. 4. Use Custom Evaluations: Leverage custom evaluations tailored to your specific use case. Generic metrics often fall short, so customize your evaluations to focus on your application's unique requirements. Lastly, continuously monitor and review the product performance. By implementing these strategies, you can significantly reduce hallucinations and enhance the reliability of your LLM-powered application. Stop nonsense output before they start – set up the right evaluations today! Sign-up for a Expert Call with one of the specialists on this topic: https://lnkd.in/eN_PpebD or Check-out LangEvals: https://lnkd.in/e9FTb6aD

Our dev team is speedy 🏎️ 💨 as well tonight. Adding GPT4o to the new playground feauture to compare your models and use the best chosen model for your AI use case. Reach out if you want to learn more. https://lnkd.in/eN_PpebD